A rapid spectrophotometric method for quantitative determination of in heated milk and milk products Ak Adhikari, D Sahai, On Mathur

To cite this version:

Ak Adhikari, D Sahai, On Mathur. A rapid spectrophotometric method for quantitative determination of lactulose in heated milk and milk products. Le Lait, INRA Editions, 1991, 71 (5), pp.555-564. ￿hal- 00929267￿

HAL Id: hal-00929267 https://hal.archives-ouvertes.fr/hal-00929267 Submitted on 1 Jan 1991

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Lait (1991) 71,555-564 555 © Elsevier/INRA Original article

A rapid spectrophotometric method for quantitative determination of lactulose in heated rnllk and rnllk products

AK Adhikari, 0 Sahai, ON Mathur

National Dairy Research Institute, Division of Dairy Chemistry, Kamal (Haryana), India 132001

(Received 20 August 1990; accepted 20 February 1991)

Summary - The reddish-brown colour produced on heating protein-free filtrate of heated milks containing less than 1.0% and free of ions in 6.4 N sulphuric acid was used for quantitative determination of lactulose. The absorptivity of the coloured solution of lactulose at 330 nm was 14- fold more than that obtained from 62.S-fold more concentration of lactose and other . By fol- lowing the optimum conditions established in this method, a minimum of 10 mg/100 ml concentration of laetulose can be determined in high heated unsweetened milk and milk products. The method was found to be very useful and sufficiently sensitive under dairy industry conditions where the ion removed step can be omitted if raw milk filtrate is taken for blank preparation. lactulose 1carbohydrate 1 heat treatment 1dried milk 1 UHT milk

Résumé - Méthode rapide de dosage par spectrophotométrie du lactulose dans le lait et les produits laitiers chauffés. La méthode de dosage du lactulose proposée est basée sur sa proprié- té à produire une coloration rouge brune au cours du chauffage (100 oC - 20 min) en milieu sulfu- rique (6,4 N). L'absorption de la solution de lactulose colorée à 330 nm est 14 fois supérieure à celle d'une solution de lactose et d'autres aldoses dont la concentration est 62,5 fois supérieure. Cette méthode permet, en suivant les conditions optimales établies, de déterminer une concentration mini- male de 10 mgl100 ml de lactulose dans des laits et produits laitiers non sucrés ayant subi un fort traitement thermique. Elle est en outre très utile et suffisamment sensible à l'échelle industrielle où l'étape de désionisation peut être supprimée si le filtrat de lait cru est pris comme blanc. lactulose 1 hydrate de carbone 1 traitement thermique 1poudre de lait liait UHT 556 AK Adhikari et al

INTRODUCTION methylamine method of Nikerson et al (1976) is not very useful in milk where the lactose content is much higher than that of Lactulose (4-0- B-D-galactopyranosyl-D- lactulose. The spectrophotometric method fructo-) is formed in heated milks of Parrish et al (1980) is time-consuming by the alkaline isomerization of lactose and requires separate determination of ail catalysed by the free amino group of case- . in (Richards and Chandrasekhara, 1960; Adachi and Patton, 1961). Lactulose oc- This paper describes a simple and rapid curs in 2 forms: as lactulose in free solu- spectrophotometric method which does not tion, and as N-deoxylactulosyl-L-Iysine require separation of lactulose from lac- which is covalently bound to the amino tose or other aldoses and can be conven- group of the milk proteins (Henry et al, iently applied for quantification of lactulose 1948). in unsweetened milk and milk products. The formation of lactulose in heated milks is dependent on time and tempera- MATE RIALS AND METHODS ture of heating and is also significantly affected by the pH of the environment (Adachi, 1959; Martinez-Castro and Ola- The principle of this method is based on the fact no, 1980). The lactose concentration in that are decomposed far more easily than aldoses by the action of minerai acids raw milk plays an important role in the for- which produces an intense red or violet coloura- mation of lactulose (Andrews, 1989). tion, while the colour appears relatively slowly Several methods have been used for and is less intense when minerai acids act upon the determination and quantification of lac- aldoses (Stanèk et al, 1963). As heated milks contain no other than lactulose (Gordon tulose in milk and milk products. The major et al, 1956), the principle applied in this method problem with quantification of lactulose is specifie and consistent. arises from the fact that the concentration of lactose is far in excess of that in milk and milk products. Most of the methods Mater/ais therefore include preliminary separation of mixtures or application of a AR grade lactulose (98.0%) was obtained from specific reaction to rem ove the lac- Sigma: Chemical Co, USA. 6.4 N H2S04 (AR tose and thereafter quantification of lactu- grade) and 40.0% TCA (AR grade) solution lose either by a general carbohydrate were freshly prepared before use. Amberlite IR- method or by TLC, GLC or HPLC meth- 120 and Amberlite IR-458 (Sigma Chemical) were used as cation and anion exchanger re- ods. Such methods are not only cumber- spectively. HCI 3 N and NaOH 2 N (AR grade) some and expensive, but also time- were used for washing the column. consuming when a large number of sam- pies have to be analysed. The spectrophotornetrlc method of Standard methods used Adachi (1965) utilizes a lengthy procedure for comparative study of oxidation of aldoses to aldonic acid by hypoiodate, whereas the anthrone method Lactulose was separated by TLC method of of Zagrodzki et al (1968) is reported to Martinez-Castro and Dlano (1981) followed by give slightly higher values as anthrone quantification by the method of Adachi (1965) reacts with 5-hydroxymethylfurfural. The and cysteine-hydrochloride-carbazole colorimet- Spectrophotometric determination of lactulose 557 ric method of Fujita and Iwatake (1931). The re- The volume of the effluent was adjusted to 100 sults obtained were compared with the H2S04 ml by removing excess water under vacuum at method. Total carbohydrate content in the milk 65-70 -c. filtrate was determined by the phenol-sulphuric acid method of Brich and Mwangelwa (1974). Development of colour

Milk samples Five ml of the above effluent was taken into a Pyrex 15 x 1.5 cm-test tube fitted with a glass UHT milk samples were obtained from a com- stopper treated with 5.0 ml of 6.4 N H2S04 and mercial milk plant (indirect type) which had been placed into a constantly boiling water bath processed at 140 °C/3 s. Sterilized (120 °C/15 (100 ± 0.5 "C) for 20 min. The tube was cooled min) and boiled milk (10 min) samples were pre- immediately to 37 ± 2 oC and absorbance was pared in the laboratory from fresh cow's milk. measured within 30 min at 330 nm by a spectro- Skim milk powders (roller- and spray-dried) were photometer (Spectronic 200) against a blank collected from the experimental dairy at the Na- prepared in a similar manner but with 5 ml of tional Oairy Research Institute, Karnal, India. 40% TCA solution in place of the heated milk fil- trate. For dried unsweetened milk powders, 1.0 9 of the sam pie was dissolved in 10 ml of Method warm distilled water, diluted to 100 ml with 40% TCA solution (dilution coefficient 10-2) and the above procedure was followed for quantification. Ten ml of warmed milk (37 oC) was treated with 40% TCA solution and kept for 10 min to precipi- tate out the total proteins, and the final volume was made up to 100 ml (dilution coefficient 10-1) Calibration curve with the same TCA solution. The whole content was filtered using Whatman No 42 filter paper. The calibration curve for lactulose was prepared by taking standard lactulose solution (conc 0.2- 1.2 mg/5 ml) and developing the colour in a si mi- Removal of interfering ions lar manner as for the milk filtrate.

To remove the interfering ions (cations Ca2+, Mg2+, Na+, K+ and anions Cit3-, HPO~+, HC0:ï, RESUL TS AND DISCUSSION SO~- and RCOO- and their complexes; Holt et al, 1981), the filtrate was passed through a cat- The action of minerai acids on led ion exchange column then an anion exchange column in free base according to the procedure to a number of decomposition products, of Adachi (1965). The cation exchange column sorne of which were coloured (Stanèk et of Amberlite IR-120 in sodium form and the an- al, 1963). Besides a certain amount of hu- ion exchange column of Amberlite IR-458 were mus-type decomposition, both aldohex- filled to 10 cm height in a 45-cm high 18-mm oses and ketohexoses yield HMF as the diameter column with glass wool and a flow con- main intermediate, which on further de- trol device. The former was treated with 100 ml composition with acid produces furfural, a of 3 N HCI and then washed with distilled water until the effluent showed a negative chloride colouring substance. test. The latter was treated with 100 ml of 2 N NaOH and th en rinsed with water until the efflu- ent showed a negative base test with phenol- Absorbance spectm (À maxJ phthalein indicator. The total milk filtrate was then allowed to pass successively through the 2 columns at the rate of 4-5 ml per min, followed Absorbance spectra of standard lactulose by washing with 100-110 ml of distilled water. solution (conc 2.92 x 10-3 molli) was ex- 558 AK Adhikari et al

amined in the range of 310-345 nm wave- of lactulose gave nearly 14 times more col- , length by a Spectronic 20 0 spectropho- ouration (00 == 0.14). This reveals that the tometer (fig 1). The absorbance maxima lactulose concentration of 62.5 times less (Âmax) was slightly broad and noted at 330 than lactose and other aldoses produces nm of 0.616 x 102 molli absorptivity. 14 times more colouration with sufficient specificity.

EffectofconcenuaUonofacid on cotour deve/opment Effect of the /ength of heaUng Ume on cotour deve/opment The effect of the strength of sulphuric acid on lactulose and other aldoses after heat- Rapid colour development was observed ing 20 min in a constantly boiling water in the initial 20 min of heating of lactulose bath (100 ± 0.5 OC) on colour development solution in 6.4 N H2S04, whereas the col- is depicted in figure 2. It was observed that our development for other aldoses was up to an acid concentration of 6.4 N, col- less (s 0.012 OD) than that for lactulose for our development for 1.0% lactose, a similar extent of heat treatment (fig 3). and solution was significantly Lactulose concentration of 0.02% produc- less (00 $ 0.01), whereas 0.016% solution es 14 times more colouration (OD == 0.178)

0.16 0.16

0.16 0.16 E gO.14 0.14 '":;; ~0.12 ~ 0.12 ] 1 ~0.10 ~ 0.10 8 8· 0.06 0.06

0.06 0.06

O.Qo\ 0.04

0.02

5.6 5.6 6.0 6.2 6.4 6.6 6.6

315 320 325 330 Concentration of H1SO. (N)

Wavelength (nm) Fig 2. Effect of strength of H2S04 on the colour development of lactulose and other aldoses. a: lactulose (0.016%), b: lactose (1.0%), c: glu- Fig 1. Absorption spectra of lactulose (2.92 x cose (1.0%), d: galactose (1.0%). 10-3 molli) in 6.4 N H2S04 at different wave- Effet de la concentration en H~04 sur l'intensi- lengths. té de la coloration produite par le lactulose et Spectre d'absorption du lactulose (2,92 x 1o-J d'autres aldoses. a: lactulose (0,016%), b: lac- mollI) dans HzS04 6,4 N, à différentes lon- tose (1,0%), c: glucose (1,0%), d: galactose gueurs d'ondes. (1,0%). Spectrophotometric determination of lactulose 559

than lactose and other aldoses at 1% con- 0.18 centration (00 s 0.012).

0.16 Calibration curve for /actu/ose 0.14 ~ li! 0.12 The plot of absorbance versus lactulose ";; concentration (0.2-1.2 mg/5 ml) was linear ~ 0.10 at 330 nm and obeyed the Beer-Lambert's 1 law in this range. The concentration of the rOB acid used and the time of heating were im- 0.06 portant determinant factors in the linearity of the curve. The higher the acid concen- 0.04 tration and the time of heating, the greater the irregularities in the colour development 0.02 observed. The regression equation devel- oped for lactulose concentration (Y = mg 10 15 20 25 30 35 = lime of heating (min) lactulose/5 ml) and absorbance (X ob- served optical density) was calculated as:

y = 28.50 X - 0.10 Fig 3. Effect of length of heating time on the co- lour development of lactulose and other aldoses. a: lactulose (0.02%), b: lactose (1.0%), c: glu- cose (1.0%), d: galactose (1.0%). cotour contribution of different a/doses Effet de la durée du temps de chauffage sur J'in- in relation to /actu/ose equiva/ent tensité de la coloration produite par le lactulose et d'autres aldoses. a: lactulose (0,02%), b: lac- tose (1,0%), c: glucose (1,0%), d: galactose Table 1 shows the relative colour develop- (1,0%). ment of lactose, glucose and galactose at

Table 1.Colour contribution of different aldoses with respect to lactulose equivalent. Absorption à 330 nm du lactose du glucose et du galactose en fonction de leur concentration dans le filtrat.

Concentration (%) Absorbance at 330 nm in the filtrate Interfering aldoses Lactose Glucose Galactose

0.50 0.000 ( 0.00)* 0.000 (0.00) 0.000 ( 0.00) 0.75 0.001 ( 0.00) 0.000 (0.00) 0.001 ( 0.00) 1.00 0.004 ( 0.18) 0.002 (0.00) 0.003 ( 0.00) 1.25 0.006 ( 1.32) 0.004 (0.18) 1 0.007 ( 1.89) 1.50 0.018 ( 8.16) 0.009 (3.03) 0.016 ( 7.02) 2.00 0.039 (20.12) 0.021 (9.89) 0.029 (14.43)

* Figures in parentheses are mg equivalent lactulose per 100 ml 01 Iiltrate. Les chiffres entre parenthèses représentent des mg d'équivalent lactulose pour 100 ml de filtrat. 560 AK Adhikari et al

Table Il. Effect of intertering aldoses on colour contribution at the same absorbance value. Poids relatif des différents aldoses pour une même valeur d'absorption.

Lactulose Absorbance Relative weight equivalent (mg/tOO ml filtrate) at330nm of aldoses (mg/tOO ml)

Lactose Glucose Galactose

10 0.02 2000 1800 20 0.04 2000 various concentration in relation to mg Effect of anion exchange resin equivalent lactulose/100 ml of milk filtrate. on reducing sugars and raw mi/k fi/trate lt is evident from the table that 1.0% of lac- tose, galactose and glucose in milk filtrate The effect of anion exchange resin on re- contribute to the similar coiour intensity of ducing sugars is presented in table V. The '" 0.18, 0.00 and 0.00 mg of lactulose per result shows that on passing 1% lactose 100 ml in the respective milk filtrates. This solution through the anion exchange resin c1earlyreveals that the intensity of colour there was a negligible increase in optical contribution by aldoses is much less than density at 330 nm. The initial value of that of the colour developed by lactulose 0.010 increased to 0.013 on passing 1% even up to the 1.0% level. Table Il repre- lactose solution, while the initial value of sents the colour contribution of mg equiva- 0.181 dropped slightly to 0.174 after pass- lent weight of lactulose in comparison to ing the lactulose solution (conc 1 mg/ml) the mg equivalent of aldoses. lt is evident through the anion exchange resin. At the from the results that '" 100, 180 and 200 same time when the raw milk filtrate was times weight equivalent of lactose, galac- tose and glucose contribute a similar inten- sity of colouration to that of lactulose on following the optimum conditions pre- Table III. Percentage of recovery of lactulose scribed in this method. added to milk/milk filtrate. Pourcentage de récupération du lactulose ajou- té au lait/filtrat de lait. Recovery studies

Lactulose Lactulose % Recovered Recovery studies were carried out by add- added recovered ing standard lactulose solution in heated (mg/Sml) (mg/Sml) milk and milk filtrate (in case of milk prod- ucts). The results obtained are presented 1.0 0.98 97.79 in table III. Appraximately 97-101% recov- 2.0 2.03 101.61 ery was observed, indicating high accura- 3.0 3.00 100.04 cy and reproducibility of the method devel- 4.0 3.91 97.83 oped. Spectrophotometric determination of lactulose 561

Table IV. Lactulose content of heated milk and milk products - comparison of 3 different methods. Teneur en lactulose de laits et produits laitiers chauffés. Comparaison de 3 méthodes différentes.

Sample Lactulose (mg/100 g)

nc method Cysteine - HCI- carbazole method H:zS04 method

Boiled milk (10 min) 84.24 (2.50)" 83.02 (2.75) 85.40 (2.55) Sterilized milk (15 min) 166.77 (2.22) 162.60 (1.80) 165.24 (2.08) UHT milk 58.60 (1.80) 57.24 (1.35) 56.04 (1.56) (indirect type) SMP (roller-dried) 478.28 (2.90) 474.30 (2.75) 479.80 (2.90) SMP (spray-dried) 266.17 (2.05) 261.40 (2.25) 268.34 (2.60)

" Figures in parentheses are standard deviations.

passed through the anion exchanger. the change in and in the nature of initial value of 0.153 dropped sharply to the solution (2%) when passed 0.028 due to removal of intertering ions through an anion exchange resin IRA-400- and the subsequent colour development OH. However, they observed a rapid ad- was very insignificant in contributing to lac- sorption of o-çlucose, D- and D- tulose value. This reveals that neither were in the hydroxide form of resin; D- the reducing sugars adsorbed nor was the fructose was transformed to o-psicose, but lactose isomerized to lactulose while pass- no evidence was found of changes in re- ing through the anion exchanger. Turton ducing sugars Iike lactose or lactulose. and Pacsu (1955) observed a negligible Adachi (1965) reported no adsorption or

Table V. Effect of anion exchange resin on reducing sugars (le lactose and lactulose) and raw milk fil- trate. Effet de la résine échangeuse d'anions sur l'absorption des sucres réducteurs (lactose et lactulose) et du filtrat de lait cru.

Standard lactose Raw milk filtrate Standard lactulose solution Added lactulose (mg/5 ml) solution (1%) (cane 1mg/ml)

Untreated Treated Untreated Treated Untreated Treated Practical Determined % loss

OD 0.010 0.010 0.155 0.030 0.185 0.178 5.0 4.86 2.80 0.005 0.D10 0.160 0.035 0.178 0.170 0.015 0.020 0.145. 0.020 0.180 0.175 Mean 0.010 0.013 0.153 0.028 0.181 0.174 562 AK Adhikari et al isomerization of lactulose while passing it same quantity of raw milk filtrate to make through strong cation and anion exchange up the total volume to 5 ml. Both the sys- resins for a long time to remove interfering tems were treated in a similar manner for ions and thereafter successfully deter- colour development as depicted in figures mined lactulose content spectrophotomet- 5 and 6. The Iinearity of both the slopes rically. Our findings also showed a negligi- curves reveals that they obeyed the Beer- ble adsorption of lactulose and no Lambert's law accurately in the given con- isomerisation of lactose to lactulose while centration range of lactulose and raw milk passing through the anion exchange resin. filtrate. This signifies the sensitivity and ac- This might be due to the strong acidic en- curacy of the method, provided the raw vironment present in resin provided by the milk filtrate is used in place of 40% TCA milk filtrate containing 40% TCA solution. solution needed for the preparation of the blank.

Omission of the ion exchange step: effect of changes in absorbance Efficacy ln the determlnatlon of tectu- with raw mi/k fi/trate and addition Jose content by uslng raw m//k fi/trate of /actu/ose solution asb/ank

To study the efficacy of the method by Table VI shows the values of lactulose omitting the ion exchange step, raw milk concentration added and determined in filtrate replaced heated milk filtrate and the raw milk filtrate when the same raw milk fil- lactulose solution was added at an in- trate concentrations were used for blank creasing concentration, (fig 4) with the preparation. The results reveal an accura-

0.18 0.24

0.22 0.18

0.20

0.18 ê 0 0.18 '"1; ê ~ 0.14 .~ 0.10 i 1ii 012 0.08 i f 0"10 0.08 ~ 0.08 8 0.08 0.04

0.04 0.02 0.02

o.oo"----~~~~~~~~~~~~ 0.2 0.4 0.6 0.8 1.0 1.2 o Raw milk filtrate (ml) Fig 4. Calibration curve for lactulose (conc 1.16 x 10-4 to 7.01 X 10-4 mol/l). Courbe de calibration pour le lactulose (conc Fig 5. Absorbance of raw milk filtrate. 1,16x 10-4-7,01 x 10-4molli). Absorption de filtrat de lait cru. Spectrophotometric determination of lactulose 563

values may be attributed towards the effect of interfering ions, lactose and aldoses, 0.55 which might have sorne influence on col- our development in this context.

0.45

0.40 E Comparlson with standard methods g 0.35

'"di 0.30 The method developed was compared with .~ the procedure described by Martinez- ~ 0.25 Castro and Olano (1981) and the cysteine- } 020 hydrochloride-carbazole method of Fujita 0.15 and Iwatake (1931). The results obtained

0.10 (table IV) by the method developed were comparable, consistent and reproducible 0.05 with sufficient accuracy (standard deviation of 2.9) as compared to the other 2 meth- 8 12 18 20 Lactuloee added (mg/5 ml) ods. The slightly higher values obtained for milk powders may be attributed to the high- Fig 6. Absorbance of laclulose solution healed er pre-heating temperatures followed un- with raw milk filtrate. der Indian conditions. Absorption de la solution de lactulose chauffée avec un filtrat de lait cru. CONCLUSIONS cy of 101.85% (mean value) in determining The rapid and simple colorimetrie method the lactulose concentration when raw milk developed can successfully be performed filtrate was taken for blank preparation in- within 2.5 to 3 h for routine quantitative de- stead of 40% TCA solution. Slightly higher termination of lactulose in unsweetened

Table VI. Lactulose content in raw milk flltrate, when raw milk filtrale was taken for blank preparation. Détermination des teneurs en lactulose ajouté dans un filtrat de lait cru (filtrat de lait cru pris comme blanc).

Vol raw milk Vollactulose Lactulose content (mg/5 ml) % Accuracy filtrate taken (ml) solution added (ml) (b/a x 100) Added(a) Determined (b)

1 4 16 16.28 101.75 2 3 12 12.44 103.66 3 2 8 8.28 103.50 4 1 4 3.94 98.50 5 0 0 0.24

Mean 101.85 564 AK Adhikari et al heated milk and milk products, and is also drates and related compounds. Anal Chem highly suitable under dairy industry condi- 28,849-863 tions where raw milk is readily available. Henry KM, Kon SK, Lea CH, White JC (1948) The ion exchange step can be omitted pro- Deterioration on storage of dried skim milk. J Dairy Res 15, 292-305 vided that the raw milk filtrate is taken for blank preparation and the colour devel- Holt C, Dalgleish DG, Jenness R (1981) Calcu- lation of the ion equilibrium in milk diffusate oped in a similar manner to the sample. and comparison with experiment. Anal Bio- Hence a large number of samples can chem113,154-163 conveniently be analysed for their lactu- Martinez-Castro l, Olano A (1980) Influence of lose content to determine the severity of thermal processing on carbohydrate compo- heat treatment with sufficient accuracy un- sition of milk: formation of epilactose. Mi/ch- der dairy industry conditions. wissenschaft35, 5-8 Martinez-Castro l, Olano A (1981) Ready detec- tion of small amounts of lactulose in dairy REFERENCES products by thin-layer chromatography. Chro- matographia 14, 621-622 Nickerson TA, Vujicic IF, Lin AY (1976) Colori- Adachi S (1959) The mechanism of the degra- metric estimation of lactose and its hydrolytic dation of lactose in strongly heated milk. products. J Dairy Sei 59, 386-390 15th Int Dairy Congr, London, 3, 1686-1691 Parrish FW, Hicks K, Doner L (1980) Analysis of Adachi S (1965) Spectrophotometric determina- lactulose preparations by spectrophotometric tion of lactulose with methylamine. Anal and high performance liquid chromatographic Chem 37,896-898 method. J Dairy Sei 63,1809-1814 Adachi S, Patton S (1961) Presence and signifi- Richards EL, Chandrasekhara MR (1960) cance of lactulose in milk products: a review. Chemical changes in dried skim milk during J Dairy Sei 44, 1375-1393 storage. J Dairy Res 27, 59-66 Andrews G (1989) Lactulose in heated milk. Bull Stanèk J, Cerny M, Kocourek J, Pacâk J (1963) tnt Dairy Fed 238, 45-52 The . Academic Press, NY, Brich GG, Mwangelwa OM (1974) Colorimetric 871-872 determination of sugars in sweetened con- Turton CN, Pacsu E (1955) The effect of anion densed milk products. J Sei Food Agric 25, exchange resins on reducing sugars. J Am 1355-1362 Chem SocTi, 1059-1061 Fujita A, Iwatake D (1931) Biochem J 242, 43 Zagrodzki S, Kulagowska A, Krol BW (1968) (cited at the 15th Int Dairy Congr; Adachi S Qualitas Plantarum Mat Veg 18, 102-112 (cit- (1959) 3, 1686-1691) ed in Andrews GR (1986) J Dairy Res For- Gordon HT, Thornburg W, Werum LN (1956) mation and occurrence of lactulose in heated Rapid paper chromatography of carbohy- milks. 53, 665-680)